UKIP are not currently predicted to win many seats. By-election results apart, in fact, they are predicted to win zero seats. Some questions have been raised about that, particularly around the question of tactical voting. UKIP might do quite well in many seats without naturally having enough support to win the seat. But if supporters of third or fourth placed parties vote tactically for UKIP, then UKIP might win the seat.
This article looks at the question of UKIP tactically voting, using data available at the time of writing. The conclusion is that tactical voting is unlikely to help UKIP, even if their support improves further, and may even hurt them.
In the 2010 general election, UKIP only gained 3.2% of the votes. But recently UKIP have gained noticeable support in the opinion polls and have been averaging around 15% over the last twelve months, and are at 17% in December 2014. Although this is higher national support than the Liberal Democrats it does not yet translate into Westminster seats. This is because, if the UKIP support is spread evenly across all the seats, then they do quite well in every seat, without ever winning one.
However it does leave them quite well placed in many seats. An interesting small table shows the frequency of (predicted) first and second placed parties across the country.
The table tells us that the most common type of seat is one where Labour is predicted to be in first place, and the Conservatives in second place. There are 179 of these seats. The next most common type is where the Conservatives are in first place and Labour is in second. There are 128 of these seats. But the next two categories are quite interesting. There are 95 seats where UKIP is in second place behind Labour, and a further 78 seats where UKIP is in second place behind the Conservatives. Together that makes 173 seats, which is a quarter of the total, where UKIP is predicted to be in second place.
The question arising is whether supporters of third and fourth placed parties will vote UKIP tactically in these seats.
Pollsters do not typically ask questions about tactical voting intentions. This makes it difficult to make scientific predictions about the likely effect of tactical voting in practice. However a recent poll by Lord Michael Ashcroft asked about potential second preference votes if an alternative vote system were being used. More details of the poll are available here, but the summary of second preferences can be expressed as a table and a transformation:
Source: Lord Ashcroft Polls
Source: Electoral Calculus
The original data from the Ashcroft poll is shown in the first table. We have performed a simple transformation on it to remove the diagonal entries so that voters second preference must be different from their first preference, and then renormalised the rows so that each row adds up to 100% (second table).
The data is quite interesting. For instance, let us start by looking at Conservative voters. Their most popular second preference is UKIP, with 40% of Conservative voters selecting UKIP. Their next most popular second preference is the Liberal Democrats, with 33%. Labour is the third most popular choice for Conservatives with 23%. So there is a definite trend amongst Conservatives to UKIP, but the other parties are also well represented.
For Labour voters however, there is a very different picture. The most popular second preference for them is the Liberal Democrats with 41%, and then other minor parties together on 34%. Both the Conservatives and UKIP are very unpopular second preferences for Labour voters, on 14% and 12% respectively.
Liberal Democrats are even less likely to switch to UKIP. Their most preferred option is to go other minor parties such as the Greens (35%), and the big two parties are also equally popular on 32%. Only 1% of Lib Dem voters nominated UKIP as a second preference.
UKIP voters themselves had the Conservatives overwhelming as their most popular second preference party (50%), with Labour trailing as next most popular (27%). Switching to the Lib Dems only received 8% support. So the antipathy between UKIP and the Lib Dems appears mutual.
A technical note of caution is that the sample size of this poll was only 1,001 people, so some of these detailed party switching estimates will be quite approximate.
We can use the results of the Ashcroft poll to estimate how voters might vote tactically in any constituency. The tactical voting model used is as follows:
If we apply this model of tactical voting to the predicted general election result (Con 30.8%, Lab 33.2%, Lib 8.2%, Ukip 17.2%, SNP in Scotland 43.4%, as at December 2014), we get an adjustment to the base result, depending on the fraction p of voters who vote tactically. The table below shows the outcomes:
|0%||251||316||19||0||45||19||Labour short 10 seats of a majority|
|10%||257||312||17||0||45||19||Labour short 14 seats of a majority|
|20%||257||314||17||0||43||19||Labour short 12 seats of a majority|
|30%||261||311||17||0||42||19||Labour short 15 seats of a majority|
|40%||266||307||16||0||42||19||Labour short 19 seats of a majority|
|50%||268||307||15||0||41||19||Labour short 19 seats of a majority|
The base case is represented by the first row, which has a tactical fraction of zero. In other words, no voter votes tactically. In this case, Labour wins 316 seats and is ten seats short of an overall majority. As the fraction of tactical voters increases to 50% (which is a realistic upper bound), we see that the situation does not change much. Labour lose about 10 marginal seats, as third-placed UKIP supporters switch to the Conservatives.
With a national support level of 17%, UKIP still wins no seats at all, even with very strong tactical voting.
We can also look at what happens if overall UKIP strength increases. Perhaps at slightly higher support levels, the tactical voting could be more relevant. The following table shows the predicted outcome of the next general election for different levels of UKIP support (with no tactical voting model yet):
UKIP really starts to win seats when its national support level gets above 25%. More details of this feature of electoral geography are described in this UKIP analysis article. We can now switch the tactical voting back on, using the high tactical fraction of 50%. The following table shows the net change in seats caused by the presence of tactical voting.
We can see that tactical voting has overall only a modest effect. The main beneficiary is the Conservatives at the expense of Labour. It also appears mildly negative for the smaller parties, including UKIP, but this result should be treated with caution. This negative UKIP effect mostly applies to Lib Dem held seats, which the strong transition model predicts will become UKIP strongholds. But in practice, those seats are probably not going to be key UKIP targets, so tactical voting may not be relevant there.
The safer conclusion is that tactical voting could benefit the Conservatives as UKIP supporters tactically vote Conservative to keep Labour out. Other tactical voting effects are smaller and less certain. However there are still two caveats even for that conclusion:
The overall conclusion is that tactical voting could have a modest postive effect on the Conservatives and against Labour at the next election, but even that is not certain. It is hard to see a definite effect on UKIP.